Dynamic feature of mitotic arrest deficient 2–like protein 2 (MAD2L2) and structural basis for its interaction with chromosome alignment–maintaining phosphoprotein (CAMP)

Mitotic arrest deficient 2–like protein 2 (MAD2L2), also termed MAD2B or REV7, is involved in multiple cellular functions including translesion DNA synthesis (TLS), signal transduction, transcription, and mitotic events. MAD2L2 interacts with chromosome alignment–maintaining phosphoprotein (CAMP), a kinetochore-microtubule attachment protein in mitotic cells, presumably through a novel “WK” motif in CAMP. Structures of MAD2L2 in complex with binding regions of the TLS proteins REV3 and REV1 have revealed that MAD2L2 has two faces for protein–protein interactions that are regulated by its C-terminal region; however, the mechanisms underlying the MAD2L2–CAMP interaction and the mitotic role of MAD2L2 remain unknown. Here we have determined the structures of human MAD2L2 in complex with a CAMP fragment in two crystal forms. The overall structure of the MAD2L2–CAMP complex in both crystal forms was essentially similar to that of the MAD2L2–REV3 complex. However, the residue interactions between MAD2L2 and CAMP were strikingly different from those in the MAD2L2–REV3 complex. Furthermore, structure-based interaction analyses revealed an unprecedented mechanism involving CAMP's WK motif. Surprisingly, in one of the crystal forms, the MAD2L2–CAMP complex formed a dimeric structure in which the C-terminal region of MAD2L2 was swapped and adopted an immature structure. The structure provides direct evidence for the dynamic nature of MAD2L2 structure, which in turn may have implications for the protein–protein interaction mechanism and the multiple functions of this protein. This work is the first structural study of MAD2L2 aside from its role in TLS and might pave the way to clarify MAD2L2's function in mitosis.

Mitotic arrest deficient 2-like protein 2 (MAD2L2), also termed MAD2B or REV7, is involved in multiple cellular functions including translesion DNA synthesis (TLS), signal transduction, transcription, and mitotic events. MAD2L2 interacts with chromosome alignment-maintaining phosphoprotein (CAMP), a kinetochore-microtubule attachment protein in mitotic cells, presumably through a novel "WK" motif in CAMP. Structures of MAD2L2 in complex with binding regions of the TLS proteins REV3 and REV1 have revealed that MAD2L2 has two faces for protein-protein interactions that are regulated by its C-terminal region; however, the mechanisms underlying the MAD2L2-CAMP interaction and the mitotic role of MAD2L2 remain unknown. Here we have determined the structures of human MAD2L2 in complex with a CAMP fragment in two crystal forms. The overall structure of the MAD2L2-CAMP complex in both crystal forms was essentially similar to that of the MAD2L2-REV3 complex. However, the residue interactions between MAD2L2 and CAMP were strikingly different from those in the MAD2L2-REV3 complex. Furthermore, structure-based interaction analyses revealed an unprecedented mechanism involving CAMP's WK motif. Surprisingly, in one of the crystal forms, the MAD2L2-CAMP complex formed a dimeric structure in which the C-terminal region of MAD2L2 was swapped and adopted an immature structure. The structure provides direct evidence for the dynamic nature of MAD2L2 structure, which in turn may have implications for the protein-protein interaction mechanism and the multiple functions of this protein. This work is the first structural study of MAD2L2 aside from its role in TLS and might pave the way to clarify MAD2L2's function in mitosis.
Vital cellular functions are highly regulated by numerous protein-protein interactions (PPIs), 2 which are essentially target-specific processes that take place via one-to-one communication. In contrast, some proteins play the role of a "hub" or "adaptor" within a PPI network by having two or more interacting partners. MAD2L2 (also known as REV7 or MAD2B), a small protein comprising just 211 amino acid residues in human and including the HORMA (Hop1, Rev7, and Mad2) domain ( Fig. 1), interacts with many partners and is crucially involved in diverse cellular functions, including translesion DNA synthesis (TLS) and mitotic events.
MAD2L2 was originally identified in studies of a yeast rev7 mutant deficient in DNA damage-induced mutagenesis by UV (1)(2)(3), where the yeast Rev7 protein was identified as a regulatory component of DNA polymerase (Pol-), an error-prone DNA polymerase specialized in TLS (4). Human MAD2L2 has been shown to interact with REV3, a catalytic component of Pol- (5), and also REV1, a binding scaffold in TLS (6,7). The crystal structure of MAD2L2 in complex with a REV3 fragment revealed that MAD2L2 interacts with REV3 through its C-terminal region, termed the "safety belt," and also implied that a second PPI site for interaction with REV1 is induced on the C-terminal ␤-sheet of MAD2L2 by REV3 binding (8). Furthermore, crystal structures of higher order complexes including MAD2L2 have clarified the role of MAD2L2 as an adaptor protein with two PPI faces that enable it to regulate the recruitment of Polto the damage site in TLS (9 -12).
In addition to its well established function in TLS, several reports have suggested that MAD2L2 has a role in mitosis. Due to its sequence similarity with MAD2 (also termed MAD2A or MAD2L1), a spindle assembly checkpoint protein that interacts with CDC20 to inhibit the activity of the APC/C ubiquitin ligase, MAD2L2 was proposed to interact with CDH1 and thereby to function as an APC/C inhibitor (13)(14)(15); however, its distinct localization in mitotic cells away from CDH1 and the phenotype caused by its depletion imply that MAD2L2 is more likely to be engaged in spindle formation and chromosome alignment (16). Indeed, the physical interaction of MAD2L2 with RAN, an important protein of bipolar spindle formation, has also been reported (17).
A zinc finger protein ZNF828 was also identified as a MAD2L2-binding protein (18,19) and was subsequently characterized as CAMP (chromosome alignment-maintaining phosphoprotein; also termed CHAMP1), a protein responsible for maintaining kinetochore-microtubule attachment in mitotic cells (19). Furthermore, it has been recently reported that CAMP is associated with a severe developmental disorder (20), and mutations in CAMP cause global development delay, intellectual disability, and dysmorphic facial features (21)(22)(23). Therefore, the structure of CAMP and its PPI has considerable interest in the context of biomedical sciences and clinical investigations.
Human CAMP consists of 812 amino acid residues containing C2H2-Zn finger domains (C2H2-ZNFs) and three characteristic repeats termed the WK, SPE, and FPE motifs (Fig. 1A); it interacts with MAD2L2 through its central region (residues 271-490) containing multiple WK motifs (19), implying that this motif has a crucial involvement in MAD2L2 binding. The WK motif is not present in the MAD2L2-binding region of either REV3 or REV1; thus, the structural basis of MAD2L2-CAMP interaction remains unknown. To date, structural studies of MAD2L2 have been limited to its complexes associated with TLS rather than those associated with its mitotic function; in addition, it is of interest to clarify the structural basis of the various PPIs of MAD2L2. In this study, therefore, the crystal structure of human MAD2L2 in complex with a CAMP fragment was determined. To our knowledge this is the first structure of an MAD2L2 complex associated with mitosis. It clarifies the mechanism underlying the MAD2L2-CAMP interaction involving the WK motif of CAMP. Furthermore, two different crystal structures indicate the dynamic nature of the C-terminal region of MAD2L2, providing direct evidence for the conformational changes of MAD2L2 together with the associated functional implications.

Structure of the MAD2L2-CAMP complex
CAMP has 13 WK sequences (Fig. 1B). Of these the region containing WK-1ϳWK-12 has been shown to interact with MAD2L2 (19). To identify the WK sequence responsible for MAD2L2 binding and to optimize the MAD2L2-binding region for the crystallographic study, we performed co-purification of MAD2L2 in complex with a CAMP fragment by immobilized metal affinity chromatography (IMAC) (Fig. 1C). Consequently, MAD2L2 in complex with a CAMP fragment containing WK-4 (residues 325-344; Fig. 1) was successfully crystallized. Crystals of human MAD2L2 in complex with the CAMP fragment were obtained in two crystal forms, Form-I and Form-II, and structures of the MAD2L2-CAMP complex in these crystals were determined at 2.1 and 2.3 Å resolution, respectively. The MAD2L2-CAMP complex adopted a monomeric structure in the Form-I crystal ( Fig. 2A), whereas unexpectedly it formed a dimeric structure through exchange of the C-terminal region of MAD2L2 and the region adopted an immature structure in the Form-II crystal (Fig. 2B). Although structural differences of MAD2L2 were observed in the two complexes ( Fig. 2C), the interactions between MAD2L2 and CAMP were essentially identical in both complexes. Therefore, the structure and interactions of the MAD2L2-CAMP complex in the Form-I crystal are described here, and the Form-II structure of the MAD2L2-CAMP complex is described below.
The crystal structure of the MAD2L2-CAMP complex comprises four ␣-helices (␣Aϳ␣D), eight ␤-strands (␤2ϳ␤8ЈЈ), and 3 10 helices (Fig. 2, A and C). As previously observed in the structure of the MAD2L2-REV3 complex (8), the CAMP fragment is wrapped in the C-terminal region of MAD2L2 termed the safety belt, which was originally noted in the MAD2-MAD1 complex (24). The structures of MAD2L2 bound to CAMP and REV3 are basically identical except for the ␤5-␣C region and the N terminus (Fig. 2D). These regions also differ between the Form-I and Form-II structures (Fig. 2C); thus, they might be correlated with dynamic features of the protein, as described below. Similar to MAD2L2, the main-chain atoms of CAMP and REV3 can be closely superimposed, indicating that overall The HORMA domain is shown by a light blue rectangle. The C-terminal region of MAD2L2 (residues 153-211), termed the safety belt, is indicated by the green line. Human CAMP comprises 812 amino acid residues (lower panel). C2H2-Zn finger domains (C2H2 ZNFs), WK, SPE, and FPE motifs are shown by blue, magenta, yellow, and green rectangles, respectively (19). The region used in this crystallographic study is indicated by the magenta line below the domain diagram together with its amino acid sequence (residues 325-344). B, WK sequences of human CAMP. Human CAMP has 13 WK sequences designated WK-1 ϳ WK-13. For clarification, tryptophan and lysine residues are colored in magenta. The WK-4 responsible for MAD2L2 interaction is highlighted in light blue. C, result of co-purification of MAD2L2-CAMP complex. The first column (CAMP) shows amino acid residues of CAMP fragment used. The second column (WK) shows WK sequence in the CAMP fragment. The third column (Complex) shows results of co-purification of the complex. The symbol ϩ indicates success in co-purification. The forth column (Crystal) shows results of crystallization. The symbols ϩ and Ϫ indicate success and failure in crystallization of the complex, respectively.

Structure of human MAD2L2/REV7 bound to a CAMP fragment
the MAD2L2-CAMP and MAD2L2-REV3 complexes adopt virtually the same structures.

Structural details of the interaction between MAD2L2 and CAMP
A previous structural study on the MAD2L2-REV3 complex revealed that multiple prolines of REV3 are crucially involved in the interaction with MAD2L2 (8) The carbonyl oxygen of Pro-1885 is supported by a hydrogen bond formed with the side chain of Tyr-37 of MAD2L2. Consistent with the structural study, a biochemical study denoted the consensus sequence for MAD2L2-binding as XPXXXPP, where and X are a hydrophobic and any residue, respectively, and the second proline is less important than the third proline (25).
Distinct from REV3, the region from residues 325 to 344 of CAMP contains the sequence "PXXXAP," where the second proline is replaced with alanine (Fig. 2D). The present structure clarifies the mechanism of interaction between MAD2L2 and CAMP (Fig. 3A). Pro-336 of CAMP is accommodated in the pocket formed by Tyr-63, Phe-169, and Trp-171, where Pro-336 is recognized by Trp-171 of MAD2L2 via a stacking inter-

Structure of human MAD2L2/REV7 bound to a CAMP fragment
action, similar to the recognition of Pro-1880 of REV3. Ala-340 and Pro-341 are accommodated in the pocket formed by Tyr-37, Glu-59, Leu-60, Tyr-63, and Phe-146, also in a similar manner to Pro-1884 and Pro-1885 of REV3. The carbonyl oxygen of Pro-341 is also supported by a hydrogen bond formed with the side chain of Tyr-37 of MAD2L2, as observed for REV3. Although the CAMP fragment lacks the second proline, Tyr-63 of MAD2L2 still interacts with the corresponding residue, Ala-340, by van der Waals interactions. Interestingly, unlike the MAD2L2-REV3 complex, a water molecule is present in the pocket and forms hydrogen bonds with the amide nitrogen of Ala-340 of CAMP and the carbonyl oxygen of Phe-146 of MAD2L2, resulting in a network of hydrogen bonds. These interactions might compensate for the lack of stacking interactions that would be provided by the second proline (Pro-1884) of REV3. Shigella IpaB also interacts with MAD2L2 (26), and its binding region similarly contains alanine instead of the second proline, suggesting that a similar network with a water molecule might be involved in the interaction between MAD2L2 and IpaB. It seems, therefore, that the second proline is not necessary but can be replaced with a small residue such as alanine so that a water molecule can mediate the interaction.
The WK motif has been identified as a novel motif in CAMP (19). The crystal structure revealed that both Trp-334 and Lys-335 of the motif are involved in binding to MAD2L2 (Fig. 3A). The aromatic side chain of Trp-334 forms van der Waals interactions with Leu-173 of MAD2L2, and the basic side chain of Lys-335 interacts with the acidic side chain of Asp-178 of MAD2L2 via ionic contacts. Furthermore, the ring of Pro-333 preceding the WK residues stacks with the aromatic side chain of His-151 of MAD2L2. In addition to the interactions described above, hydrogen bonds mediated by water molecules are involved in the interaction between MAD2L2 and CAMP (Fig. 3A).

Analysis of the MAD2L2-CAMP interaction
To pinpoint the significant interactions in the MAD2L2-CAMP complex, residues of CAMP or MAD2L2 were replaced with alanine, and the amount of CAMP fragment that co-purified with immobilized His-tagged MAD2L2 was evaluated (Fig.  3, B and C). As expected, mutation of Tyr-63 or Trp-171 in MAD2L2 markedly reduced the interaction between MAD2L2 and CAMP (lane 5 or 9 in Fig. 3B), as did mutation of Phe-169 (lane 8 in Fig. 3B), suggesting that van der Waals interactions formed by these aromatic residues of MAD2L2 are crucial for formation of the MAD2L2-CAMP complex (Fig. 3A). In contrast, mutation of Pro-336 in CAMP only slightly impaired the interaction between MAD2L2 and CAMP (lane 5 in Fig. 3C), although mutation of Pro-341 in CAMP had a large impact on the interaction, as expected (lane 8 in Fig. 3C). This indicates that the stacking interaction between Trp-171 of MAD2L2 and Pro-336 of CAMP is less important for formation of the MAD2L2-CAMP complex, in contrast to the significant contribution of the stacking interaction between Trp-171 and Pro-1880 in formation of the MAD2L2-REV3 complex (8,25).
More interestingly, mutation of Trp-334 in the WK motif of CAMP greatly reduced the interaction (lane 3 in Fig. 3C), whereas mutation of Lys-335 had only a small impact (lane 4 in Fig. 3C). Furthermore, double (W334A/K335A, W334A/ P341A) and triple (W334A/P336A/P341A) mutations markedly reduced the interaction (lane 11, 12, or 13 in Fig. 3C). These results suggest that the formation mechanism of the MAD2L2-CAMP complex differs from that of the MAD2L2-REV3 complex; namely, van der Waals interactions provided by Trp-334 and Pro-341 of CAMP are mainly responsible for the interaction with MAD2L2, whereas Pro-333, Lys-335, and Pro-336 make a moderate contribution that might be utilized for fine-tuning to increase the specificity of the interaction between MAD2L2 and CAMP. As described above, the P336A mutation in CAMP had a small effect on the interaction with MAD2L2, although the W171A mutation in MAD2L2 led to a marked reduction of binding affinity. This suggests that van der Waals interactions made by Ala-336 in CAMP mutant might be sufficient to form the MAD2L2-CAMP complex due to a large contribution from residues such as Trp-334 and Pro-341 and/or a supportive contribution from other residues.
To verify the results of the in vitro interaction between MAD2L2 and CAMP in human cells, immunoprecipitation assays using full-length CAMP were performed (Fig. 4). In good agreement with the in vitro results, Y63A, W171A, and Y63A/ W171A mutations of MAD2L2 had a substantial impact on the interaction with CAMP (Fig. 4, A and B). Furthermore, the W334A/K335A, W334A/P341A, and W334A/K335A/P341A mutations of CAMP abolished the interaction with MAD2L2 (Fig. 4, C and D). Significantly, the deletion of residues 324 -350 in CAMP also abolished the interaction (Fig. 4, C and D). These observations indicate that the structure and interaction of the MAD2L2-CAMP complex described here is likely to reflect those in human cells. Consistently, mutations in MAD2L2 had an impact on the localization of MAD2L2 in mitotic cells (Fig. 5). Wild-type MAD2L2 was co-localized with full-length CAMP in the mitotic spindle and mitotic chromosomes (Fig.  5A). Wild-type MAD2L2 was localized to the mitotic spindle and to mitotic chromosomes, whereas MAD2L2 carrying the Y63A, W171A, or Y63A/W171A mutation was not detected in mitotic chromosomes (Fig. 5B). This clearly indicates that the interactions observed in the crystal structure are essential for co-localization with CAMP at chromosomes in mitotic cells. Interestingly, MAD2L2 carrying these mutations could still localize to the mitotic spindle. This implies that MAD2L2 may have mitotic partners other than CAMP and might interact with those proteins by alternative mechanisms. Previous studies have shown that MAD2L2 interacts with spindle-related proteins such as the small GTPase RAN (17) and clathrin light chain A (CLTA) (27). MAD2L2 might retain localization in spindle by interactions with such spindle-related proteins. Further analyses will be required to explain localization of MAD2L2 in mitotic spindles. Taken together with the crystal structure and interaction assay results, these findings reveal the interaction mechanism distinct from that of MAD2L2-REV3 complex.
The structure-based interaction analysis by IMAC revealed that Trp-334 and Pro-341 of CAMP were most crucial for the binding to MAD2L2. Among 13 WK sequences of CAMP, WK-4, -7, and -8, have both corresponding tryptophan and proline residues. WK-7 and -8 have SP instead of 340 AP 341 in WK-4, implying that the polar side chain of the serine residue would be unsuitable to contact Phe-146 (Fig. 3A). In addition,  and -8 lack the proline residue preceding WK. These observations suggest that WK-4 could be determinant to interact with MAD2L2. In fact, co-immunoprecipitation assays revealed that the deletion of residues 324 -350 in CAMP, the lacking WK-4 and WK-5, abolished the interaction with MAD2L2 (Fig. 4D). WK-5 lacks AP and has 354 VS 355 instead, suggesting that WK-5 is unlikely involved in the interaction with MAD2L2. Therefore, we conclude that the CAMP fragment including WK-4 is responsible for the MAD2L2 interaction.

Dynamic feature of the C-terminal region of MAD2L2
In the Form-II crystal, the MAD2L2-CAMP complex forms a dimeric structure in which the C-terminal region of MAD2L2 is swapped between the dimers (Fig. 2B). The secondary structures of the swapped region are rearranged as follows (Fig. 2, B  and C). The ␤8Ј strand, which interacts with the swapped ␤8ЈЈ from another complex is shortened. The following ␤8ЈЈ strand is extended to the other complex and interacts with ␤8Ј and ␤5, thereby forming a ␤-sheet structure (Fig. 2B). Furthermore, the ␣E helix is newly formed between ␤8Ј and ␤8ЈЈ, whereas the ␣D helix is loosened, and the N-terminal portion of ␣C is pushed away by the ␣E; thus, the structure of the region ␤5-␣C is altered (Figs. 2, B and D, and 6), consistent with our observation of flexible features relative to the structure of the MAD2-REV3 complex (Fig. 2D).
To avoid steric hindrance from ␣C, the side chain of Phe-126 is flipped (Fig. 6). More interestingly, the swapped ␤8ЈЈ strand of the Form-II structure is shifted by two residues relative to the Form-I structure, thereby fastening of the safety belt is incomplete (Fig. 6). Namely, the C-terminal region adopts an immature structure. In more detail, three hydrophobic residues, Leu-197, Met-199, and Leu-201, of the Form-II structure are nicely in line with, respectively, Met-199, Leu-201, and Val-203 of the Form-I structure. Given these observations, the structure of the C-terminal region of MAD2L2 is inherently flexible, and the plasticity may be correlated with the mechanism of complex formation or the function of MAD2L2.

Discussion
The crystal structure of MAD2L2 in complex with a CAMP fragment was determined in two forms, Form-I and Form-II. The Form-I structure revealed that MAD2L2 binds to CAMP via its C-terminal region termed the safety belt as also observed

Structure of human MAD2L2/REV7 bound to a CAMP fragment
in the MAD2L2-REV3 complex (8). However, the mechanism underlying formation of the complex differs from that of the REV3 complex. Namely, the importance of the contribution made by each interaction varies between the two complexes; these differences might be the result of fine-turning interactions that have evolved to increase the specificity for partner proteins. Such a strategy would be reasonable for the maintenance of PPI networks connected to a hub protein, because it might reduce the impact on cellular functions caused by blocking a certain PPI by mutations or inhibitors. In other words, a structural basis for the diverse mechanisms of interaction between MAD2L2 and partner proteins might facilitate the development of specific inhibitors for MAD2L2-related PPIs.
The Form-II structure revealed the dynamic feature of the C-terminal region of MAD2L2. This could have an analogy with the spindle assembly checkpoint protein, MAD2. In humans, MAD2 shares 22% amino acid identity with MAD2L2, which on interaction with its partner proteins undergoes a marked conformational change from an open (O-MAD2) to a closed (C-MAD2) conformer whereby the C-terminal region of MAD2 moves from the edge of the ␤6 sheet toward the edge of the ␤5 sheet to bind the target (24,28). A previous study based on circular dichroism spectra suggested that MAD2L2 also undergoes a conformational change in its C-terminal region to bind REV3 (29). Although the structure of MAD2L2 without a partner protein, namely the "open" structure of MAD2L2, remains unknown, the present study provides convincing data to support a mechanism in which the C-terminal region of MAD2L2 opens and closes to bind a partner protein. Energetic studies have clarified that MAD2 functions in checkpoint activation via the formation of several types of dimer through the ␣C helix (30 -34), whereas whether MAD2L2 functions as a dimer is an open question. Recently, a second MAD2L2-binding site was identified in REV3, implying that the dimeric feature of MAD2L2 might facilitate two-site binding to REV3 to increase affinity and specificity or dimer formation might promote simultaneous interactions with multiple proteins (35). The size exclusion chromatography purification for the MAD2L2-CAMP complex used in this crystallographic study shows that the complex exists as a monomer in solution; thus, the swapped dimer in the Form-II may be an artifact in crystallization. Alternatively, the Form-II structure implies that formation of the incompletely fastened safety belt may be possible in the monomeric complex, and the immature structure might reflect an intermediate state during the partner binding. Arrangement of amino acid residues on the ␤8Ј and ␤8ЈЈ strands, namely the second PPI face, are altered (Fig. 6). REV1 binds to the ␤8Ј and ␤8ЈЈ strands of MAD2L2, and Leu-186, Gln-200, and Tyr-202 on these strands are crucial for the interaction. In the incompletely fastened safety belt, Leu-186 is located in a position similar to that in the Form-I structure or previous structures of MAD2L2 complexes, whereas Gln-200 and Tyr-202 on the ␤8ЈЈ strand have shifted positions (Fig. 6), supposedly impairing   REV1 binding. Thus, the immature structure of the safety belt would prevent the second PPI. In other words, the structure of safety belt may regulate the availability of the second PPI. This implication is consistent with the PPI model of MAD2L2 (8,10). The biological and functional relevance of the incompletely fastened safety belt and dimer formation of MAD2L2 are puzzles to be solved by further study.

Structure of human MAD2L2/REV7 bound to a CAMP fragment
In summary, we have presented the structure of MAD2L2 in complex with a CAMP fragment, thereby revealing the structural basis of the interaction between MAD2L2 and CAMP. This work is the first structural study to examine a MAD2L2 interaction not involved in TLS, and it might pave the way to clarifying the mitotic roles of MAD2L2. Furthermore, we have revealed new structural features of MAD2L2, providing mechanistic implications of PPIs by MAD2L2. Recently, biological functions of MAD2L2 in non-homologous end-joining and DNA cross-link repair have been rediscovered (36 -38). Therefore, it is becoming increasing crucial to clarify the mechanisms of PPIs engaged by MAD2L2, and these are our future goals.

Protein production and purification
The central region of human CAMP including the WK motifs (residues 271-490) is known to interact with MAD2L2 (19). To optimize the MAD2L2-interacting region for the present crystallographic study, a series of co-expression vectors derived from the pET-Duet1 vector (Novagen) and containing cDNAs of human His-tagged MAD2L2 and a human CAMP fragment varying in length from residues 271 to residues 490 were constructed and evaluated for protein production, purification, and crystallization. Consequently, MAD2L2 in complex with a CAMP fragment comprising residues 325-344 was successfully crystallized. Protein production and purification of this MADL2L2-CAMP complex are described in detail below.
The cDNA encoding human CAMP (residues 325-344) was cloned into the NdeI-XhoI site of pET-Duet1 containing the cDNA of human MAD2L2 in the EcoRI-PstI site (29), producing a vector expressing Mad2L2 with an N-terminal hexameric histidine tag and the CAMP fragment. In this study an R124A mutation was introduced in MAD2L2 to facilitate crystallization (8,29). The expression vector was used to transform Escherichia coli BL21(DE3)CodonPlus. Cells were grown at 310 K to a cell density of ϳ1.0 at 660 nm in LB medium and then cultured for a further ϳ20 h at 288 K after the addition of 0.2 mM isopropyl ␤-D-1-thiogalactopyranoside. The harvested cells were suspended in 10 ml of buffer I (50 mM HEPES-NaOH, pH 7.4, and 500 mM NaCl) per gram of cells and lysed by sonication. The cell lysate was clarified by centrifugation for 1 h at 277 K (48,300 ϫ g). Subsequent purification was carried out at 277 K. The supernatant was applied to a nickel-nitrilotriacetic acidagarose resin (Qiagen), and the resin was washed first with buffer II (50 mM HEPES-NaOH, pH 7.4, and 1.5 M NaCl) and then with buffer III (50 mM HEPES-NaOH, pH 7.4, and 100 mM NaCl). The His-tagged MAD2L2-CAMP complex was eluted stepwise with buffer III containing 50 -500 mM imidazole, diluted with buffer IV (50 mM Tris-HCl, pH 8.5), and then applied to an anion-exchange column, HiTrap Q HP (5 ml) (GE Healthcare). The bound protein was eluted with a linear gradient of 0 to 500 mM NaCl over a total volume of 95 ml. The eluted protein passed through a size exclusion column, HiLoad 16/600 Superdex 75 pg (GE Healthcare), equilibrated with buffer V (20 mM HEPES-NaOH, pH 7.4, and 100 mM NaCl) and then concentrated to 20 mg/ml using a Vivaspin (10-kDa nominal molecular weight limit) concentrator (Sartorius). The purity of the MAD2L2-CAMP complex was confirmed by SDS-PAGE with Coomassie Brilliant Blue staining. The purified protein was frozen with liquid N 2 and stored at 193 K until use.

Crystallization, data collection, and structure determination
Crystallization of the MAD2L2-CAMP complex was performed by the sitting drop vapor-diffusion method using a commercial kit to screen crystallization conditions. Drops were prepared by mixing 0.5 l of protein solution with 0.5 l of reservoir solution. Crystals were obtained under several conditions after a week at 293 K. After optimization of the conditions, crystals suitable for X-ray diffraction study were eventually grown in two crystal forms: Form-I and Form-II. Crystals of Form-I were obtained with a reservoir solution consisting of 0.1 M sodium cacodylate, pH 6.6, and 4.0 M sodium formate. Crystals of Form-II were obtained with a reservoir solution consisting of 0.3 M potassium thiocyanate and 20% PEG 3350.
Before the X-ray diffraction study, the crystals were transferred to a buffer comprising the reservoir solution plus 20% ethylene glycol for cryo-protection. Each crystal was picked up in a nylon loop and cooled and stored in liquid N 2 via an Universal V1-Puck (Crystal Positioning System, Inc.) until use. X-ray diffraction data of frozen crystals were collected under a stream of N 2 gas at 100 K on the BL-17A beamline at Photon Factory (Tsukuba, Japan) using a pixel array photon-counting detector, PILATUS3 S6M (DECTRIS). Diffraction data were integrated, scaled, and averaged with programs XDS (39) and SCALA (40).
The crystal structures of the MAD2L2-CAMP complex were determined by molecular replacement using the structure of REV7-REV3 complex (PDB ID 3ABD) as a search model with the program PHASER (41). Manual model building and fitting were performed by the program COOT (42) followed by structure refinement using the program PHENIX (43). Structure factors and final coordinates were deposited in the Protein Data Bank (PDB IDs 5XPT and 5XPU). The data collection and refinement statistics are given in Table 1. Secondary structures are defined with the program DSSP (44).

Interaction analysis by bacterial co-expression and IMAC
A cDNA encoding a human CAMP fragment (residues 271-374) was cloned into the NdeI-XhoI site of pET-Duet1 containing the cDNA of human MAD2L2 in the EcoRI-PstI site (29). Site-directed mutations in MAD2L2 or CAMP were introduced by using a PCR-based method. His-tagged MAD2L2 was co-expressed with the CAMP fragment by a procedure similar to that described above. Interaction analysis by IMAC was performed as described previously (8). In brief, the cell lysate was applied to nickel-Sepharose High Performance beads (GE Healthcare). The beads were washed first with buffer II and then with buffer III. The bound proteins were analyzed by SDS-PAGE with Coomassie Brilliant Blue staining and the

Structure of human MAD2L2/REV7 bound to a CAMP fragment
ChemiDoc Touch Imaging System (Bio-Rad). These assays were performed three times.

Interaction analysis by immunoprecipitation
The cDNA for human CAMP was cloned as described previously (19). The cDNA for human MAD2L2 was amplified by PCR from the MegaMan human transcriptome cDNA library (Agilent Technologies). The cDNAs for CAMP-truncated mutants were created by PCR. These cDNAs were subcloned into pEGFP-C1 (Clontech). The cDNAs for point mutation of CAMP and MAD2L2 were generated by using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions.
HeLa Kyoto cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. Transfection of GFP-tagged CAMP or MAD2L2 constructs was performed by using FuGENE HD Transfection reagent (Promega) according to the manufacturer's instructions. For synchronization at early mitotic phase, HeLa Kyoto cells transfected with GFPtagged MAD2L2 mutants were incubated for 12 h and then cultured in the presence of 2 mM thymidine for 24 h, released from thymidine block for 10 h, and fixed and stained.
The following rabbit polyclonal antibodies were used in immunofluorescence (IF), Western blotting (WB), and immunoprecipitation (IP): anti-CAMP (Sigma, HPA008900) diluted at 1:500 for IF and WB, and anti-GFP (Life Technologies, A11122) diluted at 1:1000 for WB and IP. The mouse monoclonal antibody against MAD2L2 (BD Transduction Laboratories, 612266) was used at a dilution of 1:500 for WB and at a dilution of 1:250 for IF.
HeLa Kyoto cells expressing GFP-tagged CAMP or MAD2L2 constructs were lysed and sonicated in lysis buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 10% glycerol, 0.5 mM EGTA, and 1 mM DTT) supplemented with phosphatase inhibitor mixture (Roche Applied Science) and protease inhibitor mixture (Roche Applied Science). After centrifugation for 10 min at 277 K at 16,000 ϫ g, the supernatants were precleared with IgG-conjugated magnetic beads (Dynabeads M-280, ThermoFisher Scientific) for 1 h at 277 K and incubated with antibody against GFP for 1 h at 277 K and then with IgG-conjugated magnetic beads for 2 h at 277 K. Precipitates were washed three times with lysis buffer, and the samples were then separated by SDS-PAGE and analyzed by immunoblotting.

Immunofluorescence and microscopy
Cells were fixed with ice-cold methanol for 10 min at 253 K after pre-extraction using PHEM buffer pH 7.0 containing 60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgSO 4 , and 0.1% Triton X-100 for 1 min at room temperature, blocked with phosphate-buffer saline containing 3% BSA and 0.01% Triton X-100 at room temperature for 30 min, and then incubated with primary antibodies overnight at 277 K. Goat anti-rabbit IgG Alexa-Fluor 488 and goat anti-mouse Alexa-Fluor 568 (Molecular Probes) were used as secondary antibodies. Immunofluorescence images were obtained using a Personal DV microscope (Applied Precision) equipped with a Plan Apochromat 100ϫ oil objective lens (NA 1.40) (Olympus) mounted on an inverted microscope (CoolSNAP HQ; Photometrics), driven by softWoRx software (Applied Precision, LLC).